Font Size: a A A

Research On Key Technologies Of Evidence Modeling Based On Open Framework D-S Theory

Posted on:2012-03-14Degree:MasterType:Thesis
Country:ChinaCandidate:B C WangFull Text:PDF
GTID:2218330362952707Subject:Circuits and Systems
Abstract/Summary:PDF Full Text Request
Evidence theory is an effective method of uncertainty reasoning, and widely used in situation assessment, pattern recognition, fault diagnosis and other areas. Evidence modeling is the first step to reason and make a decision, but the modeling method was not given in evidence theory. Evidence modeling has been one of the bottlenecks restricting evidence theory to be widely used currently. Open Framework D-S Theory (OFDST) extended evidence theory whose framework becomed complete with evidence combination, expands the scope of application of evidence theory, and provides an effective method that can represent and combine evidences when a complete identification framework can not been obtained. The surpport function, objecting function, and potential function establish a uniform scale for qualitative and quantitative evaluation of different evidences. To solve the key problems of the evidence modeling, the main contributions are as follows:Firstly, the latest theory and application advances of some key problems of evidence modeling are summarized systematically, and the solutions are analyzed.Secondly, based on the statistical characters, a method of evaluating sensor credibility is presented. The data simulation results show that this method is feasible and effective, and can truly reflect the credibility of information sources.Thirdly, a modeling method is proposed based on sensor credibility. To solve the problem that the current evidence model does not or not fully consider the credibility, the sensor credibility is integrated into the evidence model by plausibility discount strategy. By the analysis of Open Framework D-S Theory, because the credibility of various information sources is considered and the evidence is modeled under OFDST, the evidence model is more reasonable, and the risk of the decision is also reduced.Fourthly, the problem of modeling and combination of different recognition framework is presented, and a normalization method is proposed.Finally, in order to make further research, other prospects of evidence theory are presented.
Keywords/Search Tags:data fusion, evidence theory, Open Framework D-S Theory, evidence modeling, evidence credibility, plausibility function discount, identification framework
PDF Full Text Request
Related items